Structure of pre-miR-31 reveals an active role in Dicer–TRBP complex processing

Significance Mature microRNAs, in complex with Argonaute proteins, function to control protein levels in the cell. Mature microRNAs are produced via a series of enzymatic processing events from longer primary and precursor transcripts. While proteins are commonly regulators of various steps in the microRNA biogenesis pathway, no proteins that interact with pre-miR-31 have been identified. Therefore, the mechanism by which miR-31 levels are controlled has until now remained elusive. The present study reveals the three-dimensional structure of pre-miR-31 and uncovers a mechanism by which processing by the Dicer–TRBP complex is regulated internally by the pre-miR-31 structure. Insights into the structure and molecular determinants of microRNA biogenesis have implications in RNA-targeted drug development and design of short hairpin RNAs for gene silencing.

5ʹ and 3ʹ extensions were included to facilitate library preparation and sequencing.These regions were not predicted to disrupt the folding of pre-miR-31.b and c) Fraction of mutations and deletions on a per-residue basis upon reaction with 2.5% DMS (b) or 5% DMS (c).A control where DMSO was included in the reaction rather than DMS indicates minimal background mutations (black bars in b and c).Secondary structure was rendered using RNA2Drawer (6) Fig. S2.Oligo controls for NMR chemical shift assignment of FL-pre-miR-31.Four oligonucleotide controls were designed to cover the entire FL-pre-miR-31 sequence.Non-native tetraloops (red) were included to cap oligos that truncated the apical loop.Secondary structures were rendered using RNA2Drawer (6).Fig. S3.Assigned chemical shifts of TopA RNA.a) 1 H-13 C HMQC and b) 1 H-1 H NOESY of TopA RNA.The signals assigned to GAGA tetraloop are colored red.NMR spectra were recorded at 0.4 mM RNA concentration, 50 mM K-phosphate buffer, pH=7.5, 1 mM MgCl2 and 100% D2O.c) Sequence analysis and validation of TopA chemical shift assignments.Nucleotide numbering, sequence, and secondary structure in Vienna format for TopA RNA.NMRViewJ chemical shift prediction software was used to validate proton (H6/H8, H5/H2, H1´, H2´, H3´) and carbon (C6/C8, C2) assignments.Assigned atoms are represented with blue circles (open and closed), while grey boxes denote atoms that are not present in a given base.Deviation from the predicted chemical shift is shown with the offset from the center.Filled circles indicate that there are chemical shifts for atoms with the same set of attributes in the BMRB.Open circles indicate atoms that have a prediction, but for which no exact matches of the attributes are available in the BMRB.Secondary structures were rendered using RNA2Drawer (6). 1 H-13 C HMQC and b) 1 H-1 H NOESY of Top RNA.c) 1 H-1 H NOESY spectrum overlay of fully-protiated (pink) and A H C H -labeled (teal) Top RNA.Secondary structure is colored to indicate the proton position (teal) in the A H C H -labeled Top RNA sample.All other sites are perdeuterated (black).NMR spectra were recorded at 0.4 mM RNA concentration, 50 mM K-phosphate buffer, pH=7.5, 1 mM MgCl2 and 100% D2O.d) Sequence analysis and validation of Top chemical shift assignments.Nucleotide numbering, sequence, and secondary structure in Vienna format for Top RNA.NMRViewJ chemical shift prediction software was used to validate proton (H6/H8, H5/H2, H1´, H2´, H3´) and carbon (C6/C8, C2) assignments.Assigned atoms are represented with blue circles (open and closed), while grey boxes denote atoms that are not present in a given base.Deviation from the predicted chemical shift is shown with the offset from the center.Filled circles indicate that there are chemical shifts for atoms with the same set of attributes in the BMRB.Open circles indicate atoms that have a prediction, but for which no exact matches of the attributes are available in the BMRB.      in the absence (black) and in the presence (magenta) of 4.8 mM paramagnetic compound Gd(DTPA-BMA).NMR spectra were recorded at 0.48 mM RNA concentration, 50 mM K-phosphate buffer (pD=7.5), 1 mM MgCl2 and 100% D2O at 37 °C and 800 MHz.The H8-C8 assignments are labeled on the spectra.b) FL pre-miR-31 secondary structure colored to indicate the position of junction residues G29, A40 and A41 (red, green, and purple, respectively).Residues shaded gray were not included in the analysis.c) sPRE data for aromatic H8 protons of FL pre-miR-31.The errors of the sPRE values were obtained from the linear regression as described previously (7).The average sPRE values ± one standard deviation for unpaired (blue shading) and paired (grey shading) residues are indicated.

4 Fig. S1 .
Fig. S1.In vitro DMS-MaPseq of pre-miRNA-31 RNA.a) Construct design and reactivity scores.5ʹand 3ʹ extensions were included to facilitate library preparation and sequencing.These regions were not predicted to disrupt the folding of pre-miR-31.b and c) Fraction of mutations and deletions on a per-residue basis upon reaction with 2.5% DMS (b) or 5% DMS (c).A control where DMSO was included in the reaction rather than DMS indicates minimal background mutations (black bars in b and c).Secondary structure was rendered using RNA2Drawer(6)

Fig. S5 .
Fig. S5.Deuterium labeling improves spectral quality by reducing overlap.a) Overlay of the unlabeled (fully protiated, gray) and A 2r G r -labeled (blue) pre-miR-31 1 H-1 H NOESY spectra.b) Chemical structures of the four nucleosides indicating sites of selective deuteration (gray shade).Sites containing non-exchangeable protons are colored blue.c) Overlay of the unlabeled (fully protiated, gray) and A H C H -labeled (teal) pre-miR-31 1 H-1 H NOESY spectra.d) Chemical structures of the four nucleosides indicating sites of selective deuteration (gray shade).Sites containing non-exchangeable protons are colored teal.e) Overlay of the unlabeled (fully protiated, gray) and G H U 6r -labeled (purple) pre-miR-31 1 H-1 H NOESY spectra.f) Chemical structures of the four nucleosides indicating sites of selective deuteration (gray shade).Sites containing non-exchangeable protons are colored purple.

Fig. S6 .
Fig. S6.Summary of the secondary structure, NOE connectivity, and chemical shift assignment validation for FL pre-miR-31.a) The secondary structure is shown beneath the sequence in Vienna format along with arrows to denote helical regions.b) NOE upper limit restraints for specified proton pairs used in CYANA and AMBER calculations are drawn as black bars.The thickness of the bar is representative of the strength of the measured NOE.c) NMRViewJ chemical shift prediction software was used to validate assignments of H6/H8, H5/H2, H1ʹ, H2ʹ, H3ʹ, C8 and C2.Protons that have been assigned in FL pre-miR-31 are indicated with blue circles (open and closed).Deviation from the predicted chemical shift is represented by deviation from the center.Predictions that are based on examples in the database of chemical shifts are shown as filled circles, predictions without data are shown as open circles.Filled grey squares are present in for nucleotides that do not contain a given proton or carbon.Unassigned resonances have open grey symbols.

Fig. S8 .
Fig. S8.Imino region of 1 H spectra of pre-miR-31 FL at different temperatures.The NMR spectra were recorded at 0.3 mM RNA concentration, 50 mM K-phosphate buffer, pH=7.5, 1 mM MgCl2 and 90% H2O/10% D2O at 600 MHz and at temperatures between 5 and 37 °C as indicated on the right side of spectra.

Fig. S9 .
Fig. S9.pH-dependence of unpaired adenosines.a) BEST selective 1 H-15 N HSQC spectra of 15 N-AU labeled FL pre-miR-31, collected at various pH conditions.b) Secondary structure of FL pre-miR-31.c) Quantification of chemical shift perturbations (pH=5.8-pH=7.5).A33 and A58 were not included in this analysis.The cross-peak of A33 is too broad to detect at pH=5.8 and cross-peak of A58 is severely overlapped at pH=5.8.Coloring follows the secondary structure in panel b.

Fig. S10 .
Fig. S10.Solvent paramagnetic relaxation effect analysis of FL pre-miR-31 reveals solvent accessibility in the loop region.a) 1 H-13 C HSQC spectra of 15 N/ 13 C A,G-labeled FL pre-miR-31 in the absence (black) and in the presence (magenta) of 4.8 mM paramagnetic compound Gd(DTPA-BMA).NMR spectra were recorded at 0.48 mM RNA concentration, 50 mM K-

Fig. S14 .
Fig. S14.Correlation plot between measured and back-calculated RDCs for the lowest energy pre-miR-31 structure.

Fig. S17 .
Fig. S17.Structure of pre-miR-31 is not sensitive to the divalent cation.The NMR spectra were recorded at 0.01 mM RNA concentration, in 24 mM K-phosphate buffer, pH=7.5, 5 mM MgCl2 or 5 mM CaCl2 (as indicated on the left) and 90% H2O/10% D2O at 10°C and at 600 MHz.

Fig. S18 .
Fig. S18.Dicer processing co-factor, TRBP, inhibits pre-miR-31 processing.a) Secondary structures of pre-miR-31.b) Denaturing polyacrylamide gel resolving the RNA products upon incubation with Dicer.c) Denaturing polyacrylamide gel resolving the RNA products upon incubation with Dicer-TRBP complex.d) Quantification of pre-miR-31 processing with either Dicer (closed circles) or the Dicer-TRBP complex (open circles).Average and standard deviation from n=3 independent assays are presented.

Fig. S19 .
Fig. S19.Mismatches in the stem of pre-miR-31 do not significantly impact Dicer-TRBP processing of the substrate RNA.Processing assays for WT pre-miR-31 and a) G14U, b) C18U, c) A54G, and d) C18A RNAs.Quantification of pre-miR-31 processing with the Dicer-TRBP complex.Average and standard deviation from n=3 independent assays are presented.Regions of the secondary structures of constructs designed to stabilize or destabilize the stem mismatches are included for clarity.Mutations are indicated with red lettering.

Fig. S20 .
Fig. S20.Mismatches in the stem region have no impact on Dicer processing.a) Secondary structures of constructs designed to stabilize or destabilize the stem mismatches.Mutations are indicated with red lettering.b) Dicer processing assay of pre-miR-31 RNAs.Average and standard deviation from n=3 independent assays are presented.

Fig. S21 .
Fig. S21.Structure at the dicing site serves as a control element for Dicer processing.a) Predicted secondary structures of constructs designed to minimize the internal loop at the dicing site.Mutations are indicated with red lettering.b) Minimization of the internal loop at the Dicing site does not inhibit Dicer processing.c) Dicing site mutants with expanded internal loop structures.Mutations are indicated with red lettering.d) Pre-miR-31 RNAs with larger internal loops at the Dicer cleavage site have reduced Dicer processing, relative to WT.

Fig. S22 .
Fig. S22.Dicer processing assays for apical loop mutations.a) Secondary structures of pre-miR-31 RNAs with smaller apical loops.Sites of mutation are denoted with red lettering.b) In vitro Dicer processing assays reveal a significant reduction in substrate cleavage for G32C and G32C/A33C RNAs.c) Secondary structures of mutants designed to extend the pre-miR-31 apical loop.Insertions are indicated with red lettering.d) Dicer processing of pre-miR-31 RNAs with larger apical loops was moderately reduced relative to WT.Average and standard deviation from n=3 independent assays are presented.

Fig. S23 .
Fig. S23.A two base-pair junction between the apical loop and dicing site recovers reduced Dicer processing efficiency due to large apical loop size.a) Secondary structures of WT, AP+2, 40UUG, and AP+5 pre-miR-31 RNAs which have 8, 10, 11, and 13 nucleotide apical loops, respectively.Mutations are indicated with red lettering.b) Dicer processing assay of pre-miR-31 RNAs.Average and standard deviation from n=3 independent assays are presented.

Table S1 .
Chemical shift completeness.a a "/" indicates a given atom is not present in the nucleoside.

Table S2 .
NMR restraints and structural statistics for the FL pre-miR-31 structure.a (8,9)tistics are reported for the entire structure unless otherwise specified.bStatistics for the 20 structures with lowest target function.cRMSD: root mean squared deviation.dStatistics for the 20 lowest energy structures.eThe 20 amber-refined structures were evaluated using the MolProbity webserver(8,9)

Table S3 .
SEC-SAXS data acquisition, sample details, data analysis, model fitting, and software used.
a Average and standard deviation from n=3 independent assays are presented.b P value based on an unpaired parametric t-test.c * P value <0.05, ** P value <0.01, *** P value <0.001, **** P value <0.0001, NS indicates no significant difference.

Table S6 .
Thermal stability of pre-miR-31 RNAs.Tm values were obtained by fitting CD thermal denaturation profiles to a two-state unfolding model using sloping baselines.Average and standard deviation from n=3 independent assays are presented.

Table S7 .
Synthetic DNA templates and associated RNA constructs.Italicized nucleotides correspond to the sequence complementary to the T7 promoter.c Red nucleotides indicate non-native tetraloop sequences. b

Table S9 .
DNA primers for generation of the pre-miR-31 (NMR) template.

Table S10 .
Amplification primers for template.